Dynamic Monitoring of Cellular Remodeling Inducedby the Transforming Growth Factor-β1

Andrea Staršíchová, Lukáš Kubala, Eva Lincová, Zuzana Pernicová,Alois Kozubík, and Karel Souček

Abstract

The plasticity of differentiated adult cells could have a great therapeutic potential, but at the same time, it ischaracteristic of progression of serious pathological states such as cancer and fibrosis. In this study, we reporton the application of a real-time noninvasive system for dynamic monitoring of cellular plasticity. Analysis ofthe cell impedance profile recorded as cell index using a real-time cell analyzer revealed its significant increaseafter the treatment of prostate epithelial cells with the transforming growth factor-β1. Changes in the cell indexprofile were paralleled with cytoskeleton rebuilding and induction of epithelial–mesenchymal transition andnegatively correlated with cell proliferation. This novel application of such approach demonstrated a greatpotential of the impedance-based system for noninvasive and real-time monitoring of cellular fate.

Key words: real-time cell analysis, cell plasticity, epithelial–mesenchymal transition, transforminggrowth factor-β1, F-actin, cytoskeleton remodeling.

1. Introduction

Thephenomenonof plasticity of differentiated adult cells could have agreat therapeutic potential, but at the same time, it is characteristic ofprogression of serious pathological states. Epithelial–mesenchymaltransition (EMT) is a crucial process in embryogenesis, but it alsooccurs during progression of tumors derived from epithelial cells(for review, see (1 )). The transforming growth factor-β1 (TGF-β1)is an important growth factor inducing remodeling of epithelial cells.TGF-β1 induces a complex change of the gene expression profile,which leads to the induction of cell cycle arrest, increased cell

Shulin Li (ed.), Biological Procedures Online, Volume 11, Number 1© to the author(s) 2009DOI: 10.1007/s12575-009-9017-9 URL: springerprotocols.com; springerlink.com

316

migration, and spreading (2–4 ). In general, determination of thequality andquantity of remodelingof epithelial cells is a complex issue.It usually includes quantification of expression of epithelial andmesenchymal markers (E-cadherin, N-cadherin, and vimentin),visualization of cytoskeletal rebuilding (F-actin), migration, andinvasive assay (wound healing and migration through Matrigelmatrix; (5 )). Conventionally, most of the approaches mentionedare basedon a time-consuming end-point analysis of the state ofwholecell populations combined with advanced techniques of analysis ofindividual cells with the use of flow cytometry or digital microscopictechniques and image analysis. However, neither the episodic northe spatial resolution of these techniques is capable of registering verysmall and fast changes in cellular morphology. Currently, label-freeand noninvasive methods based on electronic cell sensor arrays weresuggested for themonitoring of cell physiology, particularly adhesion,spreading, and transient changes in cell morphology (6–9 ). Towidelyaccept this methodological approach and to correctly and preciselyinterpret data for these measurements is crucial to obtain precisecorrelation with cell morphology and overall phenotype using arelevant reference method. However, well-described modelsapplying this methodological approach with different cell lines andvarious cell plasticity modulating conditions are missing. Here, weshowed that the impedance-based real-time cell analyzer (RTCA)allows dynamic monitoring and quantification of cell remodelingduring TGF-β1-induced EMT in non-transformed prostate epithelialcells. This novel application of such approach demonstrated a greatmedium-throughput potential of the impedance-based system fornoninvasive and real-time monitoring of cellular fate.

2. Materialsand methods

2.1. Cells BPH-1 cells were obtained from the German Collection ofMicroorganisms and Cell Cultures and cultivated in RPMI1640, supplemented with 20% bovine fetal serum (both PAA),5µg/ml transferrin, 5 ng/ml sodium selenite, and 5µg/ml insulin(Invitrogen). The cell lines were cultivated in Nunc (ThermoFisher Scientific) cultivation dishes, flasks, and plates in a humidifiedincubator at 37°C in an atmosphere of 5% CO2.

2.2. Real-time cellimpedance analysis

AceaE-plates®96were used for noninvasive real-timemeasurementwith the use of an xCELLigence RTCA SP system including RTCASoftware version 1.1 (both Roche). First, a standard backgroundmeasurement was performed using 100 μl of complete cultivationmedia. BPH-1 cells were trypsinized, quantified, and seeded in

Dynamic Monitoring of Cellular Remodeling Induced by TGF-β1 317

additional 100 μl of cultivation media in a final concentration of30,000 cells per cm2. The cells were monitored continually every1 min in the first 45 min after the seeding and then every 1 h for aperiod of 96 h. Recombinant TGF-β1 (Millipore) treatment withvarious concentrations in triplicate was performed 24 h after theseeding of the cells. Formation of contractile microfilaments wasblocked by cytochalasin B (CB), Helminthosporium dematioideum(Calbiochem) dissolved in methanol (MeOH). The cells were pre-treated with TGF-β1 (10 ng/ml) for 68 h and treated with CB(10 μg/ml) for another 3 h. The cells were monitored continuallyevery 15 s after the CB addition. In this case, data are presented asa normalized cell index (CI; normalized at the time of 68 h).Cultivation of the cells and their treatment were performed understandard conditions (37°C/5% CO2).

2.3. Cell counts The numbers of trypsinized BPH-1 cells in the culture weredetermined using a Coulter Counter® ZM (Beckman-Coulter).

2.4. ATP assay Intracellular ATP was detected in BPH-1 cells by the commercialATP cellular kit (Biothema, Sweden). The cells were incubatedaccording to the experimental procedure, the supernatant wasremoved, and the cells were lysed by the Somatic cell ATP releasingreagent (Sigma-Aldrich). Then, 50μl of lysate wasmixedwith 20μlof ATP reagent containing D-luciferin, luciferase, and stabilizers.Intracellular ATP contents were determined using a microplateluminometer LM-01T (Immunotech).

2.5. Fluorescentand light microscopy

F-actin was visualized after the staining of paraformaldehyde (2%)fixed and permeabilized BPH-1 cells with phalloidin-fluoresceinisothiocyanate (Sigma-Aldrich) using a fluorescent microscope(Olympus IX-70, Fluoview II CCD camera). Nuclear counter-staining was performed by using 4′,6-diamidine-2′-phenylindoledihydrochloride (DAPI; Fluka). Cell morphologywas documentedby phase contrast on the same microscope.

2.6. Western blot BPH-1 cells were treated by various concentrations of TGF-β1 fordifferent time intervals and harvested in radioimmunoprecipitationassay buffer (50 mM Tris–HCl pH7.4, 1% NP-40, 0.25% sodiumdeoxycholate, 150 mMNaCl, protease inhibitor cocktail, and phos-phatase inhibitor cocktail set II (Merck)). Protein concentration wasdetermined using detergent-compatible protein assay (Bio-Rad).The cell lysates were sonicated (5 s, Sonifier® B-12, BransonUltrasonics Corp), spun, and mixed with 3× sodium dodecyl sulfate(SDS) loading buffer (240 mM Tris–HCl pH6.8, 6% SDS, 0.02%bromphenol blue, 30% glycerol, 3% β-mercaptoethanol). Equivalentquantities of protein (20µg) were separated by SDS-polyacrylamidegel electrophoresis and transferred onto polyvinylidene fluoridemembranes (Millipore) using established procedures. The

318 A. Staršíchová et al.

membranes were blocked in Tris-buffer saline (20 mM Tris–HClpH7.2, 140 mM NaCl) containing 0.1% Tween 20 and 5% non-fat milk. The levels of phosphorylated (Ser465/467) and totalSmad2, and expression of vimentin, a characteristic mesenchymalmarker, were analyzed with specific primary antibodies (CellSignaling and Sigma-Aldrich). Anti-β-actin (A5441) was fromSigma-Aldrich; horseradish peroxidase-conjugated anti-mouseIgG (#NA931) and anti-rabbit IgG (#NA934) were from GEHealthcare. Detection of antibody reactivity was performed usingImmobilon Western HRP Substrate (Millipore). Densitometricmeasurements were performed using ImageJ software (NIH)and normalized to the expression of β-actin.

3. Resultsand discussion

Data acquisition demonstrated a linear increasing of the CI valuesin control cells during the time interval observed. However, thislinear trend was significantly changed by TGF-β1 in less than12 h after the treatment in a concentration-dependent manner(Fig. 1a). Concentrations of 1 and 10 ng/ml induced a significantsteep increase in CI values, which reached a plateau in 48 h. Theimpedance-based determination is by its nature dependent on thenumber of adherent cells. Thus, we compared CI determinationwith the analysis of cell numbers to clarify the contribution ofchanges in cell numbers and the morphological alternation of cellsto detected CI values. In parallel with the E-plates® 96, the cellswere seeded at the same density on 40 mm dishes and 4-well plates,and treated with TGF-β1 in the same experimental design. Atvarious time intervals after the treatment, the numbers of trypsinizedcells were determined using a Coulter Counter simultaneously withthe determination of metabolically active viable cells based on deter-mination of intracellular ATP in cell lysate. Our data, shown inFig. 1b, c, demonstrate thatTGF-β1 induced antiproliferative effectsin BPH-1 cells in a time- and concentration-dependent manner.These trends are in negative correlation withCI values acquired withthe use of RTCA. Taken together, these data showed that theTGF-β1 induced antiproliferative effects in BPH-1 cells are paralleledby an increase of cell impedance.

It is a well-known fact that when stimulated with TGF-β1,epithelial cells undergo EMT and exhibit a significant formationof actin stress fibers that emanate from focal adhesions (10 ). Inthis case, our results demonstrate the usual uncertainty of imageanalysis which is limited by confluence of the cells. The controlcells reached a relatively high density after 72 h of cultivation.

Dynamic Monitoring of Cellular Remodeling Induced by TGF-β1 319

Based on the F-action visualization and cell morphology analysis,there is no significant difference between control cells and cells treatedwith 0.1 ng/ml of TGF-β1. However, it is evident that TGF-β1 at 1and 10 ng/ml concentrations induced formation of F-actin stressfibers and increased cell spreading (Fig. 2). These results positivelycorrelate with the activation of Smad-dependent signaling andEMT by TGF-β1 (Fig. 3). Our results showed transient phosphory-lation of Smad2 induced by both 1 and 10 ng/ml of TGF-β1, whichis followed by massive induction of vimentin expression. TGF-β1 at a

Fig. 1. The antiproliferative effect of transforming growth factor-β1 (TGF-β1) is associated with theincrease of the cell index (impedance) in BPH-1 cells. a TGF-β1 induces an increase of the cell index(impedance) acquired with the use of an xCELLigence real-time cell analyzer system in a time- andconcentration-dependent manner. However, at the same time, TGF-β1 strongly inhibits cell prolifer-ation quantified by counting of the cells using a Coulter Counter (b), and by ATP assay (c). The cellswere treated with various concentrations of TGF-β1 at the time marked as time zero.

320 A. Staršíchová et al.

0

0.1

1

10

hoursTGF-β1ng/ml

4824 72

Fig. 2. The effect of transforming growth factor-β1 in BPH-1 cells is associated with remodeling ofthe cytoskeleton. F-actin was visualized with the use of phalloidin-fluorescein isothiocyanate conju-gate. The nuclei were counterstained with DAPI. Images of cellular morphology were taken underphase contrast.

Dynamic Monitoring of Cellular Remodeling Induced by TGF-β1 321

concentration of 0.1 ng/ml did not induce vimentin expressionand only slightly induced phosphorylation of Smad2. Furthermore,we wanted to examine the influence of formation of stress fiberson TGF-β1-induced increase of CI values. CB, a potent inhibitorof the formation of contractile microfilaments, induced inhibitionof F-actin formation in both control and TGF-β1-treated BPH-1cells (Fig. 4a). This inhibition was paralleled by a rapid decreaseof normalized CI values in both groups (Fig. 4b). However, thedrop of CI values induced by CB was much stronger in the case ofTGF-β1-pretreated BPH-1 cells, with high abundance of F-actinstress fibers in comparisonwithCB-treated control cells. These resultsdemonstrate that TGF-β1-induced increase of CI values is at leastpartially dependent on stress fiber formation. We can summarize thatTGF-β1-induced EMT in BPH-1 cells is associated with inhibition ofproliferation, rebuilding of cytoskeleton, formation of stress fibers,and increase of spreading. These cellular changes are associated witha significant increase inCI values analyzed by theRTCA system.Until

Control

TGF-β1

P-Smad2

hours

Smad2/3

Vimentin

β-actin

8

6

4

2

O.D

. (fo

ld o

f con

trol

)

0

10

1520

3025

5

O.D

(fo

ld o

f con

trol

)

0

0.1 ng/ml 1 ng/ml 10 ng/ml

Fig. 3. Transforming growth factor-β1 induces phosphorylation of Smad2 and epithelial–mesenchymaltransition in BPH-1 cells. The cells were lysed using radioimmunoprecipitation assay buffer, sodiumdodecyl sulfate-polyacrylamide gel electrophoresis, and immunoblotting detection of p-Smad2(Ser465/467); Smad2 and vimentin was performed. Detection of β-actin served as control of equal load-ing. Densitometric measurements were performed using ImageJ software (NIH, Bethesda, MD, USA)and normalized to the expression of β-actin. The bar graph represents the average optical density.

322 A. Staršíchová et al.

now, most of the applications of RTCA focused on analysis ofadhesion or spreading were designed to study immediate interactionof cells with substrate in relatively short time intervals (6, 11 ).Here, our data suggest a novel application and a great potential ofreal-time impedancemeasurement for dynamicmonitoring of cellularremodeling and plasticity, induced after adhesion of the cells to thesubstrate and formation of stress fibers in longer time intervals.Furthermore, these measurements also show a potential misinterpre-tation of the data in experimental setups using impedance-basedmethods for the detection of cell proliferation without simultaneousmonitoring of cellular morphological changes.

1.2

1.0

0.8

0.6

68 69 70

hours

ControlTGF-β1cytochalasin B

TGF-β1 + cytochalasin B

norm

aliz

ed c

ell i

ndex

71 72

Control

MeOH

CB

TGF-β1a

b

Fig. 4. Inhibition of formation of contractile microfilaments by cytochalasin B (CB) leads to a rapid dropof cell index values. BPH-1 cells were pretreatedwith transforming growth factor-β1 (10 ng/ml) for 68 hand treated with CB (10 μg/ml) for another 3 h. a F-actin was visualized by phalloidin-fluoresceinisothiocyanate conjugate after 3 h of CB and/or vehicle treatment. b CB induces a rapid decreaseof the normalized cell index (datawere normalized at the time of 68 h) acquiredwith the use of an xCEL-Ligence real-time cell analyzer system. The arrow indicates the time interval of CB addition.

Dynamic Monitoring of Cellular Remodeling Induced by TGF-β1 323

Acknowledgment

The authors would like to thank Viktor Horváth (Roche CzechRepublic) for his assistance with xCELLigence RTCA, IvaLišková, Jaromíra Netíková and Petra Jelínková for superb technicalassistance, and Ladislav Červený for the correction of English. Thiswork was supported by grant number 204/07/0834 of the CzechScience Foundation, and grants numbers AV0Z50040507 andAV0Z50040702 of the Academy of Sciences of theCzechRepublic.

References

1. Kalluri R (2009) The basics of epithelial-mesenchymal transition. J Clin Invest 119(6):1420–1428

2. Heldin C-H, Landström M, Moustakas A(2009) Mechanism of TGF-[beta] signalingto growth arrest, apoptosis, and epithelial-mesenchymal transition. Curr Opin Cell Biol21(2):166–176

3. Pachernik J, Soucek K,Hampl A,HofmanovaJ, Kozubik A (2001) Transforming growthfactor-beta1 induces junB mRNA accumula-tion, G1-phase arrest, and pRb dephosphory-lation in human leukemia HL-60 cells. FoliaBiol 47(1):32–35

4. Soucek K, Pachernik J, Kubala L, VondracekJ, Hofmanova J, Kozubik A (2006) Trans-forming growth factor-beta1 inhibits all-transretinoic acid-induced apoptosis. Leuk Res 30(5):607–623

5. Zeisberg M (2009) Biomarkers for epitheli-al-mesenchymal transitions. J Clin Invest119(6):1429–1437

6. Xi B, YuN,Wang X, Xu X, Abassi YA (2008)The application of cell-based label-freetechnology in drug discovery. Biotechnol J3(4):484–495

7. Otto AM, Brischwein M, Grothe H,Motrescu E,Wolf B (2003)Multiparametricsensor chips for chemosensitivity testing ofsensitive and resistant tumor cells. RecentResults Cancer Res 161:39–47

8. Keese CR, Bhawe K, Wegener J, Giaever I(2002) Real-time impedance assay to followthe invasive activities of metastatic cells inculture. Biotechniques 33(4):842–844, 846,848–850

9. Spegel C,Heiskanen A, LarsHenrik, SkjoldingD, Emnéus J (2008) Chip based electroanalyt-ical systems for cell analysis. Electroanalysis 20(6):680–702

10. Maeda M, JohnsonKR,WheelockMJ (2005)Cadherin switching: essential for behavioral butnot morphological changes during an epitheli-um-to-mesenchyme transition. J Cell Sci 118(5):873–887

11. Yu N, Atienza JM, Bernard J, Blanc S, ZhuJ, Wang X, Xu X, Abassi YA (2006) Real-time monitoring of morphological changesin living cells by electronic cell sensorarrays: an approach to study G protein-coupled receptors. Anal Chem 78(1):35–43

324 A. Staršíchová et al.